Complete result settings for Ad dataset (table columns are sortable)
The experiments are described in Rooshenas and Lowd, Discriminative Structure Learning of Arithmetic Circuits, AIStats 16

Split Penalty	Standard Deviation	L1 Penalty	Initial Evidence	Train CLL	Validation CLL	Test CLL	Node#	Edge#	Feature#	Learning Time
10	0.5	0.5	10	-4.603199	-5.183774	-5.041980	42617	56009	14966	69785.494991
5	0.5	0.5	10	-4.603199	-5.183774	-5.041980	42617	56009	14966	70905.591711
2	0.5	0.5	10	-4.603199	-5.183774	-5.041980	42617	56009	14966	70508.642056
2	0.5	2	20	-5.363259	-5.977065	-5.750118	48807	64675	17442	74109.708611
10	0.5	2	20	-5.363259	-5.977065	-5.750118	48807	64675	17442	75308.835316
5	0.5	2	20	-5.363259	-5.977065	-5.750118	48807	64675	17442	75615.447704
5	0.5	1	20	-4.732752	-5.323691	-5.165703	48807	64675	17442	75627.287904
10	0.5	1	20	-4.732752	-5.323691	-5.165703	48807	64675	17442	75656.723429
2	0.5	1	20	-4.732752	-5.323691	-5.165703	48807	64675	17442	75327.201523
2	0.5	1	1	-5.122315	-5.699607	-5.529763	21034	25831	6314	75920.809282
2	0.5	0.5	20	-4.417680	-5.037826	-4.892823	48807	64675	17442	76182.058566
5	0.5	0.5	1	-4.895553	-5.537620	-5.329596	20867	25559	6266	76941.583101
5	0.5	0.5	20	-4.417680	-5.037826	-4.892823	48807	64675	17442	79351.911675
2	0.5	0.5	1	-4.982997	-5.649007	-5.408453	20936	25674	6286	79350.251927
10	0.5	0.5	20	-4.417680	-5.037826	-4.892823	48807	64675	17442	80337.650820
10	0.5	0.5	1	-4.874837	-5.538482	-5.337787	20910	25655	6262	81070.979336
2	0.5	0.1	1	-5.588159	-6.381103	-6.088897	21769	27562	6344	82655.473457
5	0.5	0.1	1	-5.618046	-6.355475	-6.096042	21358	26661	6296	83513.577006